surface temperature dynamics in response to …
TRANSCRIPT
94
Journal of Civil Engineering, Science and Technology Volume 11, Issue 2, September 2020
SURFACE TEMPERATURE DYNAMICS IN RESPONSE TO
LAND COVER TRANSFORMATION
Syed Riad Morshed, Md. Abdul Fattah, Asma Amin Rimi and Md. Nazmul Haque* Department of Urban and Regional Planning, Khulna University of Engineering & Technology, Khulna University of
Engineering & Technology, Khulna-9203, Bangladesh.
Date received: 26/04/2020 Date accepted: 20/08/2020
*Corresponding author’s email: [email protected]
DOI: 10.33736/jcest.2234.2020
Abstract — This research assessed the micro-level Land Surface Temperature (LST) dynamics in response to Land Cover
Type Transformation (LCTT) at Khulna City Corporation Ward No 9, 14, 16 from 2001 to 2019, through raster-based analysis
in geo-spatial environment. Satellite images (Landsat 5 TM and Landsat 8 OLI) were utilized to analyze the LCTT and its
influences on LST change. Different indices like Normalized Difference Moisture Index (NDMI), Normalized Difference
Vegetation Index (NDVI), Normalized Difference Buildup Index (NDBI) were adopted to show the relationship against the
LST dynamics individually. Most likelihood supervised image classification and land cover change direction analysis shows
that about 27.17%, 17.83% and 4.73% buildup area has increased at Ward No 9, 14, 16 correspondingly. On the other hand,
the distribution of change in average LST shows that water, vacant land, and buildup area recorded the highest increase in
temperature by 2.720C, 4.150C, 4.590C, respectively. The result shows the average LST increased from 25.800C to 27.150C in
Ward No 9, 26.840C to 27.230C in Ward No 14 and 26.870C to 27.120C in Ward No 16. Here, the most responsible factor is
the transformation of land cover in buildup areas.
Copyright © 2020 UNIMAS Publisher. This is an open access article distributed under the Creative Commons Attribution-Non-commercial-ShareAlike 4.0
International License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Keywords: Land Cover Change, LST Dynamics, NDBI, NDMI, Raster Based Approach
1.0 INTRODUCTION
The enormous population growth has resulted in the increase of new cities, industries, vehicles, and
infrastructures. Urbanization and industrialization have been influenced in the drastic change of land
cover types around the world. This has resulted in the extensive environmental changes around the world.
Global warming, changes in ecosystems, climate change, greenhouse emissions, increase of sea level
and many other problems are increasing day by day due to the massive transformation of land cover
types [1]. The study aimed to identify how significantly the Land Surface Temperature (LST) responses
to the change of Land Use Patterns (LUP) in some contiguous wards of a developing city. Environmental
changes such as temperature, rainfall, LST and so on does not changes equally in all portions of the city
because the development of change of land cover types do not changes equally in all directions. The city
shows the overall different changes on a large scale. The study focused on identifying the micro levels
(ward wise) change of LST and LUP of Khulna city. For these three adjacent wards, Ward No 9, Ward
No 14 and Ward No 16 were selected in this study.
In Growth of the population has been increasing the need for new housing, workplace, education, food,
and other basic needs along with the increase of human activities. Again, various activities are going on
to improve the living standard of the people, which is exerting a negative influence on Land Cover Type
(LCT). Human activities on a large scale are incessantly influencing in the massive transformation of
LCT [2]. A huge percentage of water body vacant land, agricultural land, vegetation, wetlands are
decreasing every day and transforming into buildup areas around the world. This massive Land Cover
Type Transformation (LUTT) affecting the Surface Energy Budget (SEB) through altering the climate.
One of the important parameters for estimating SEB is Land Surface Temperature (LST) through
assessing land use and land cover (LULC) pattern changes [3]. Various methods and approaches are
introduced by many researchers to measure and analyze the LST [4,5,6,7].
95
Due to urbanization and unplanned development resulted in a malignant loss of greeneries, vacant space,
waterbodies, eco-system, and ecologically sensitive habitats [8]. Many houses, shopping malls,
buildings, infrastructures are being built in every year. As a result of the unplanned urbanization, the
sensitive areas are destroyed. People are building infrastructures wherever they wish, without thinking
about the mother nature. As a result, the environment is deteriorating its balance. The earth’s average
temperature and the concentration of CO2 in the air is increasing, but the total average rainfall is
decreasing. Since 1880, the global annual average temperature rose at an average rate of 0.07°C (0.13
°F) per decade [9]. According to Shahid [10], the mean temperature in Bangladesh has risen by
0.1030C/year during the period 1958-2007 and for Khulna the rate was 0.0070C/year during 1960-2012.
But with the due to massive environmental change the mean temperature increasing rate stood at
0.041620C/year in Khulna during 2003 to 2018 [11]. If the land use transformation continues in current
rate, within 2050 the temperature will rise to 10C in Khulna and there will be no land to transform into
build area.
Increase of buildup area reduces the heat reflection capacity of soil and soil starts to retain heat which
increases the surface temperature. Destruction of waterbodies like river, ponds, wetlands, canals etc. and
increase of buildup area prevents water from entering the deep soil. For which ground water level is
declining and soil moisture retention capacity is decreasing. influencing in the decrease of Normalized
Difference Moisture Index (NDMI) value [12] and rise of land surface temperature. Again, due to the
destruction of greeneries and vegetation areas, the value of Normalized Difference Vegetation Index
(NDVI) [13] decreases and increase of buildup area increases Normalized Difference Buildup Index
(NDBI) value [14]. These three factors response mostly with the change of land use pattern and
influences in the change of LST in any area and global LST rose from 0.60C to 0.90C (1.10F to 1.60F)
during the year 1906 to 2005[11,15].
Many researchers have worked on the change of LST in Dhaka City in the literature [16,17,18,19].
Among all these researches, [16] calculated all the factors and indexes related to the change of the LST.
Sharmin [20] showed the relationship between NDVI and LST. Kafy et. al. illustrated the land cover
change and LST change in the Rajshahi City Corporation area in the literature [28]. These all research
showed the influences of land use change in the change of Land surface temperature in a city or in a
division that means in a large scale. But, the change of LUP does not occur uniformly over the city. LUP
change is much higher in industrial areas of a city rather than in the residential areas and the vegetation
areas. Again, the surrounding areas of the industrial areas have higher environmental temperatures rather
than other areas of a city. So, the calculation of the change of LUP and LST in city scale does not show
the accurate image of LST and LUP change of that city. In this study, the identification of LUP and LST
change in the ward basis attempted to solve the problems in large scale LUP and LST change analysis.
Again, the ward wise linear regression analysis between LST and NDVI, LST and NDWI, LST and
NDBI has been conducted for both years. Determination of LST changes based on LUP and surface
temperature change direction based on land use change direction has fulfilled many previous research
limitations.
The rise of LST in Dhaka city has observed 3.20C during the year 2002 to 2014 [18] and in case of
Chittagong metropolitan area it is observed 3.810C [20]. The change of average LST at the research study
area was 3.430C which is found almost same in the case of Dhaka, Chittagong, and Rajshahi City [21].
Due to the implementation of Padma Bridge and various development projects, communication system
with Khulna city and other cities of the country will be developed. This will create new investment
opportunities and therefore, the policy makers hope that the social, physical, and economic condition of
Khulna will improve a lot. As a result, most of the areas of Khulna City are hoping to be turned into
buildup areas. But unplanned urbanization, destruction of eco-friendly areas and high land cover
transformation rate will lead to various environmental problems including the rise of LST and ecological
imbalance in the city. As a result, living in this city will begin to unsuitable day to day. Various problems
will increase including decrease of soil fertility, increase of soil salinity, lowering ground water levels
96
etc. which will increase the suffering of the residence by reducing food production, increasing water
crisis, and changing environment. This research will provide information to the people about the effect
of unplanned development and unplanned land cover change on environmental change. Also inspire for
further research on the micro level environmental change due to land use and land cover changes in
Bangladesh.
2.0 METHODOLOGY
2.1 STUDY AREA
The study area was selected as three different wards, Ward No 9 Mujgunni, Ward No 14 Boyra and Ward
No 16 Choto Boyra under Khulna City Corporation area (Figure 1). These three wards are located along
the middle of KCC. In each ward, there are industrial areas, commercial areas along with residential areas.
The total area of the study area is 2113.53 acres with a population of 87430(BBS,2011). About 73% land
use type across the study area is residential and most of these are in ward no 14 & 16. After the
introduction of Mujgunni Residential Area, many buildings are being constructed at ward no 9. That is
why the land cover change has been observed much higher in this area rather than in the surrounding
wards. Although these three study areas are located side by side, but the maximum land use purposes of
these three wards were different. Whereas most of the land cover in Ward No 9 was agricultural land,
residential land in Ward No 16, buildup and open space in Ward No 14. In addition, the percentage of
mixed land use areas in these wards are much higher than other wards of KCC [10,29]. That is why these
three wards have been selected to find the micro level land surface temperature change with the
conversion of land use type.
Figure 1. Location map of study area (KCC Ward No. 9, 14, 16). (Source: KCC, 2019)
97
2.2 DATASET PREPARATION
The study has been conducted by using two satellite images of less than 10% land and scene cloud
coverage, collected from US Geological Survey (USGS). Two satellite images of 2001(march) and
2019(march) were taken to avoid the impact of seasonal temperature variations on the land surface
temperature variations in the Landsat images. And due to less possibility of rainfall in Bangladesh during
March, the accuracy or acceptability of LST result is higher than other months. The satellites specification
utilized in the study was described in Table 1.
Table 1. Satellite Sepcification.
Satellite/
Sensor
Pixel
Size
Spectral Resolution Resolution Path and
Row
Date
LANDSAT 5 30 Multispectral (6 Bands) 30 m 138/44 17/03/2001
LANDSAT 8 30 Multispectral (11 Bands) 30 m 137/44 28/03/2019
2.1 LAND USE AND LAND COVER MAP PREPARATION
In order to make land cover map, image enhancement tool is used to classify and select the area of interest.
Contrast Enhancement, Saturation, hue, intensity and Density Slicing etc. are the most used ways to
enhance Landsat images. In this study, Landsat images were enhanced by generating composite band
combination [16]. The radiometric and atmospheric correction of the Landsat satellite images has been
conducted. In this study, the generation of composite band combinations such as natural color composite,
true color composite, false color composite etc. are used to identify the land use type in the study area.
Blue, green, red and near infrared bands were used for Landsat 5 TM images of 2001 and Landsat 8 OLI
images of 2019 to find true color during data processing in Earth Resources Data Analysis System
(ERDAS) Imagine 2014.
The land cover type has been classified into five categories (buildup area, vacant land, agricultural land,
waterbodies, greeneries) in ArcGIS (geo-spatial planning tool) and EARDAS Imagine 14 to identify the
land cover pattern at the study area in the year 2001 and 2019. Then the land cover type change direction
during the study period 2001 to 2019 has been analyzed in ArcGIS 10.6 version.
2.4 ACCURACY ASSESSMENT
During the same time frame, a comparison was made between the 400 random sampling points and their
corresponding point on Google Earth images to verify the LULC classifications [20]. The most likelihood
supervised image classification method used for this research is of good accuracy as the validity rates are
more than 90% for the study area.
2.5 LAND SURFACE TEMPERATURE (LST)
Land surface temperature has been calculated for the year 2001 and 2019 in the study area by using
Landsat 5 TM and Landsat 8 OLI satellite images, respectively. Rahman and Kabir stated the massive
increase of population, industrial areas, and alternation of LCT mostly in the Khulna city since the year
2000 [30] for which the study period is chosen between 2001 to 2019. Then LST map has been generated
for both year and change in LST have been identified for each ward during the study time. Equation 1-5
were used to estimate the LST [21].
Lλ = Aλ + ML ∗ QCAL (1)
where
Lλ = TOA Spectral Radiance (W/ (m2 × sr × μm))
ML = Radiance multiplicative scaling factor for the band
98
AL = Radiance additive scaling factor for the band
QCAL = the quantized calibrated pixel value in Digital Numbers (DN)
In the second step the TOA spectral radiance (Lλ) values are converted into another variable called At-
Satellite Brightness Temperature (TB)
TB =k2
lnk1
L1+1
(2)
Where
TB =At-Satellite Brightness Temperature, in Kelvin (K)
K1, K2 = Thermal conversion constants for the band
Finally, the TOA Brightness Temperature will be converting to land surface temperature values (LST)
using the formula.
𝐿𝑆𝑇 = [𝑇𝐵/ (1 + (𝜆 ∗ 𝑇𝐵/ 𝛼)) ∗ 𝑙𝑛 𝜀] (3)
Where
LST = Land Surface Temperature in Kelvin (K)
λ = the wavelength of emitted radiance
α = hc/k (1.438 × 10-2 mK)
h = Planck constant (6.626 × 10-34 J s-1), and c = velocity of light (2.998 × 108 m s-1)
k = Boltzmann constant (1.38 × 10-23 J K-1)
ε = surface emissivity, calculated according to equation 4
𝜀 = 0.004 ∗ 𝑃𝑣 + 0.986 (4)
Where
Pv is the vegetation proportion calculated following equation 5.
𝑃𝑦 = [𝑁𝐷𝑉𝐼−𝑁𝐷𝑉𝐼 𝑚𝑖𝑛
𝑁𝐷𝑉𝐼−𝑁𝐷𝑉𝐼 𝑚𝑎𝑥]
2 (5)
To obtain the land surface temperature values in Celsius (°C), 273.15 was extracted from the initial values
(K).
2.6 NORMALIZED DIFFERENCE VEGETATION INDEX (NDVI)
The NDVI is calculated using near-Infrared spectral band and Red spectral band of Landsat’s multi-
spectral sensor. [21]. The concept of estimating the NDVI is based on the absorption of radiation in the
red (R) spectral band and the maximum reflection of radiation in the near-infrared (NIR) spectral band.
𝑁𝐷𝑉𝐼 = 𝑁𝐼𝑅 𝐵𝑎𝑛𝑑−𝑅𝐸𝐷 𝐵𝑎𝑛𝑑
𝑁𝐼𝑅 𝐵𝑎𝑛𝑑+𝑅𝐸𝐷 𝐵𝑎𝑛𝑑 (6)
NDVI values vary from −1 to + 1, with values close to −1 suggesting lack of vegetation and values close
to + 1 indicating dense vegetation.
2.7 NORMALIZED DIFFERENCE BUILDUP INDEX (NDBI)
Built-up Index (BU), Normalized Difference Built-up Index (NDBI), Index based Built-up Index (IBI)
etc. are used to analyze the built-up area. In this study, NDBI [12] is used to detect the built-up area in
the study area.
𝑁𝐷𝐵𝐼 = 𝑆𝑊𝐼𝑅−𝑁𝐼𝑅
𝑆𝑊𝐼𝑅+𝑁𝐼𝑅 (7)
99
NDBI values vary from −1 to + 1, with values close to −1 suggests no built up area and values close to +
1 indicates the dense built-up area.
2.8 NORMALIZED DIFFERENCE MOISTURE INDEX (NDMI)
NDMI is used for the analysis of water bodies in any area. The index uses remote sensing images with
Green and Near infrared bands. NDMI can efficiently enhance water information in most cases [22]. The
result of NDMI indicates the intensity of waterbodies in the area and illustrates the presence of moisture
in soil [29]. The higher value indicates higher intensity of waterbodies and the lower NDMI value
indicates absence of waterbodies.
𝑁𝐷𝑀𝐼 = 𝑁𝐼𝑅−𝑆𝑊𝐼𝑅
𝑁𝐼𝑅+𝑆𝑊𝐼𝑅 (8)
NDMI values vary from −1 to + 1, with values close to −1 suggests lack of waterbodies and values close
to + 1 indicates presence of large waterbodies.
With the calculated values of NDVI, NDMI and NDBI for both years, the NDVI map, NDMI map and
NDBI map were generated and then the change of these indexes in each ward during the study period has
been identified. Linear regression analysis has been conducted among LST and NDVI, LST and NDWI,
LST and NDBI to see which type of land use pattern is mostly responsible for the change of LST in the
study area.
2.9 RELATIONSHIP BETWEEN LST AND LAND COVER CHANGE
Several Studies [17,18,23,24] have shown that quantifying the LST's effects and spatial patterns involves
a clear understanding of the LST's relationship with different types of land cover. These studies only
explained the change in surface temperature because of the vegetation land cover transformation. In this
study the change of surface temperature due to all types of land cover transformation and their influences
has been analyzed. The study extracted the average, maximum and minimum LST values of the five types
of land cover in the year 2001 and 2019 and identified the average LST change with change rate during
the study period. The study also explored the change of average LST in different land cover, before and
after the land cover transformation. The overall study procedure followed in this study is visualized
through a methodological framework in Figure 2.
100
Figure 2. Methodological framework of the study
3.0 ANALYSIS AND INTERPRETATION
3.1 LAND USE AND LAND COVER (LULC) PATTERN CHANGE ANALYSIS
Increase of population and human activities, either on a large scale or small scale are continuously
reducing the waterbody, vegetative cover, vacant land, and agricultural land cover of the earth’s surface.
GIS based satellite image classification in EARDAS Imagine 2014 for the year 2001 and 2019 in Figure
3 shows that the study area has faced a huge amount of land cover transformation in the last 20 years. The
area of each land cover type between 2001 and 2019 for every ward was calculated through maximum
likelihood image classification and presented in Table 3. The positive sign in Table 3 indicates the
increase of land cover percentage while negative sign indicates the decrease of land cover areas. Table 3
shows that there was a drastic and rapid transformation of waterbody at Mujgunni KCC Ward No 9.
Approximately 103.76 acres of waterbodies were transformed into vacant land and buildup area and 229.4
acres of buildup area was increased at Mujgunni. Introduction of Mujgunni Residential Area has
influenced mostly in this rapid increase of the buildup areas at Ward No 9. Table 3 shows the highest
percentage (54.33%) of LULC at ward 9, then 35.65% in ward 14 and 27.44% at ward 16.
Atmospheric & Radiometric
Correction of LANSAT 5 TM &
LANDSAT 8 OLI
LU Classification (Maximum
Likelihood Supervised Image
Classification
Land Use and Land Cover Map
(2001 and 2019)
Land cover change direction
NDMI, NDVI, NDMI
map (2001 and 2019)
Spectral Radiance Model TM
Band for 2001 and Split Window
Algorithm (TIRS) Band 10 & 11
for 2019
Land Surface Temperature Map
(2001 and 2019)
Different Land Surface
Temperature Change
LST dynamics in different land cover conversion
Subdivision of
obtained value into
several ranges
Linear Regression and
Trend Analysis
Identification of
Responsible Land Cover
type for the rise of LST
Measuring LST,
NDMI, NDVI,
NDBI values
101
Figure 3. Land Use and Land Cover map of KCC (A) Ward No 9 in 2001 (B) Ward No 9 in 2019 (C) Ward No 14 in 2001
(D) Ward No 14 in 2019 (E) Ward No 16 in 2001 and (F) Ward No 16 in 2019.
Table 3. Land cover area in the year 2001 and 2019 in each ward (Unit: Percentage)
Figure 3C, 3D and Table 3 shows that about 17.83% of buildup area has increased and 5.79% of
vegetation and 1.68% of waterbody area has been decreased at KCC ward no 14. Table 3 illustrates that
buildup area was increased mostly in Ward No 9 by 27.17% and due to the filling of waterbodies and
wetlands, waterbody land cover area percentage was declined by 12.29% during the study period.
Although the rate of Land Use Transformation (LUT) was higher in KCC Ward no 9 and 14, it is noticed
less in case of ward no 16. From the field survey it is found that, due to the establishment of a residential
area in Ward No 9, many houses and building were built in there. As a result, people built many
infrastructures in there for which they filled up waterbodies, wetlands, and agricultural lands. GIS based
satellite image classification in Table 3 and Figure 3E and 3F shows that only 4.73% (28.52 acres) of
buildup area has increased, 0.70% (4.20 acres) of waterbody area and 4.76% (28.7 acres) of vegetated
area was also increased in ward 16. But about 13.72% (72.8 acres) of agricultural land has been decreased
during the study period.
Land Cover
Type
Ward N 9 Ward No. 14 Ward No. 16
2001 2019 Change 2001 2019 Change 2001 2019 Change
Agricultural 14.89% 4.11% -
10.78%
23.17% 13.65% -9.52% 29.02% 15.30% -
13.72%
Build up
Area
7.71% 34.88% 27.17% 27.31% 45.14% 17.83% 29.97% 34.70% 4.73%
Vacant Land 16.18% 12.83% -3.35% 15.52% 14.68% -0.83% 18.44% 21.99% 3.54%
Vegetation 36.56% 35.83% -0.74% 31.20% 25.41% -5.79% 21.65% 26.40% 4.76%
Waterbody 24.65% 12.37% -
12.29%
2.80% 1.13% -1.68% 0.92% 1.61% 0.70%
Total 100% 100% 54.33% 100% 100% 35.65% 100% 100% 27.44%
102
The land use and land cover (LULC) change direction analysis in ArcGIS shows a remarkable change of
LULC in 20 years. Figure 4 shows that about 110.73acres (5.24%) of waterbodies, 227.27acres (10.75%)
of agricultural land has been transformed into build up areas. The land cover change direction analysis
illustrated in Figure 4 shows that waterbody and agricultural land cover were worst affected in the study
area. Due to the high percentage of agricultural and waterbody land cover area in the study area, these
two land cover types have become transformed mostly. Moreover, many waterbodies are still being filled
for further construction of buildings in the future. As a result of urban expansion and growth of population,
vegetation, waterbodies, agricultural land, and vacant land surfaces exposed.
Figure 4. LULC change direction analysis between 2001 to 2019 at study area.
3.2 LST DYNAMICS IN RESPONSE TO NDMI
The value of NDMI lies between -1 to 0 to +1. Higher value NDMI indicates the higher density of water
surface and lower density of buildup area. Using equation 8, NDMI profile of the study area in both study
period was calculated and presented in Figure 5.
Figure 5. Normalized Difference Moisture Index map of KCC (A) Ward No 9 in 2001 (B) Ward No 9 in 2019 (C) Ward No
14 in 2001 (D) Ward No 14 in 2019 (E) Ward No 16 in 2001 and (F) Ward No 16 in 2019.
-5.24%
-0.76%
-0.59%
-10.75%
17.82%
Waterbody
Vegetation
Vacant Land
Agricultural land
Buildup Area
103
With the growth of population and because of urbanization, 5.24% (110.73 acres) of waterbodies (Figure
4) has been transformed across the study and resulted in the decrease of NDMI value (Figure 5). The
higher value of NDMI is decreased from 0.43 to 0.33 in 9 No Ward (Figure 5A, B) and increased from
0.29 to 0.27 in 16 No Ward (Figure 5E, F). But the percentage of dense water cover areas across the study
area has reduced.
As in Ward No 16, vegetation and waterbodies area has increased (4.76% and 0.70% respectively in Table
3) and buildup area increasing rate was low, the total percentage of areas with higher NDMI value was
increased in 2019 (Figure 5E, 5F). Though the higher value of NDMI at 14 No Ward has increased but
the percentage of buildup area has increased (Figure 2C, D). Therefore, it can be concluded that NDMI
has been decreasing and the percentage of areas with lower NDMI value are increasing day by day across
the study area.
Figure 6. Linear correlation between LST change in response to NDMI value change in the year (A) 2001 and (B) 2019.
From the regression analysis between LST and NDMI in Figure 6 shows an increasing trend of LST with
the decrease of NDMI value. The regression analysis for the year 2001 in Figure 6 represents the negative
relationship between LST and NDMI values i.e. decrease of waterbodies influences in the increase of
surface temperature. The value of coefficient of determination (R2) was calculated 0.9976 for the year
2001 and 0.9909 for the year 2019 which indicates the significant relationship between NDMI and LST.
Figure 6 states that the surface temperature increased during the study period because of the reduction of
the total amount of waterbodies in the study area.
3.3 LST DYNAMICS IN RESPONSE TO NDVI
The values of NDVI vary between -1 to +1. The values of NDVI more than 0.3 indicates the healthy
vegetation, the value from 0.01 to 0.02 and from 0.02 to 0.03 indicates non vegetated area and unhealthy
vegetation [29]. The NDVI values for the study area was calculated by using equation 6 and illustrated in
Figure 7. Figure 7 visualizes the declination of the healthy vegetation and increase of the unhealthy
vegetation area in every ward. NDVI change analysis shows that the percentage of the unhealthy
vegetation area and non-vegetated area has been increased at Mujgunni mostly. Where the total amount
of healthy vegetation areas declined mostly at ward no 14 showed in Figure 7C, 7D. That is why the
average LST increased mostly at ward no 9 and ward no 14 mostly which presented in Figure 7.
y = -19.177x + 23.39
R² = 0.9976
15
17
19
21
23
25
27
29
-0.3 -0.2 -0.1 0 0.1 0.2 0.3
LS
T (
0C
)
NDMI
a
y = -10.027x + 27.321
R² = 0.9909
20
22
24
26
28
30
32
-0.4 -0.2 0 0.2 0.4 0.6
LS
T
(0C
)
NDMI
b
104
Figure 7. Normalized Difference Vegetation Index map of KCC (A) Ward No 9 in 2001 (B) Ward No 9 in 2019 (C) Ward
No 14 in 2001 (D) Ward No 14 in 2019 (E) Ward No 16 in 2001 and (F) Ward No 16 in 2019.
In 2001, NDVI range was between -0.0082 to +0.48, -0.0034 to +0.44 and -0.024-0.43 at ward no 9, 14
and 16, respectively. But the value has been changed and showed a positive change condition in 2019
though, the average NDVI value declined from 0.25 to 0.1837. Transformation of vegetation areas and
agricultural land is responsible for the increase of LST in the study area. If the deforestation and cut of
urban trees are not stopped, then this situation will continue to be worse day by day.
The linear regression trend analysis of NDVI (Figure 8) for the year 2001 and 2019 shows the strong
negative correlation between LST and NDVI value. Figure 8 indicates that surface temperature decreases
with the increase of NDVI values that means with the increase of dense vegetation areas. The coefficient
of determination value found 0.986 for the year 2001 and 0.9955 for the year 2019 and both indicates the
strong or significant response of surface temperature with the change of NDVI values.
Figure 8. Linear correlation between LST change in response to NDVI value change in the year (A) 2001 and (B) 2019.
(Source: Author, 2020)
y = -11.658x + 29.465
R² = 0.986
22
24
26
28
30
32
-0.2 0 0.2 0.4 0.6 0.8
LS
T (
0C
)
NDVI
a
y = -21.172x + 28.77
R² = 0.9955
18
20
22
24
26
28
30
0 0.1 0.2 0.3 0.4 0.5
LS
T
(0C
)
NDVI
b
105
3.4 LST DYNAMICS IN RESPONSE TO NDBI
Increase of buildup areas by 376.59 acres over the study period has increased the NDBI value across the
study area. NDBI have indices value ranges from -1 to 1. The higher value of NDBI has increased from
0.056 to 0.43 at ward no 9 (Figure 9A, B), 0.10 to 0.33 at ward no 14 (Figure 9 C, D) and 0.14 to 0.30 at
ward no 16 (Figure 9 E, F). The analysis shows that NDBI has increased 8 times at ward no 9, 3 times at
ward 14 and 2 times at ward 16.
Figure 9 Normalized Difference Buildup Index map of KCC (A) Ward No 9 in 2001 (B) Ward No 9 in 2019 (C) Ward No 14
in 2001 (D) Ward No 14 in 2019 (E) Ward No 16 in 2001 and (F) Ward No 16 in 2019.
Figure 9 visualizes a huge change of NDBI during the year 2001 to 2019 across the study area.
Figure 10. Linear correlation between LST change in response to NDBI value change in the year (A) 2001 and (B) 2019.
y = 9.8639x + 26.975
R² = 0.9867
22
24
26
28
30
32
-0.4 -0.2 0 0.2 0.4 0.6
LS
T (
0C
)
NDBI
a
y = 19.25x + 23.021
R² = 0.9979
18
20
22
24
26
28
30
-0.2 -0.1 0 0.1 0.2 0.3
LS
T (
0C
)
NDBI
b
106
As the total amount of increased buildup area was most at Mujgunni, the value of NDBI is changed mostly
in this area than other wards. In 2001 the NDBI range was between -0.35 and 0.14 across the study area,
which is gradually increased to between -0.43 to + 0.43 (Figure 9). The higher value increased about 3
times and the lower value of NDBI has decreased very slightly. Therefore, it can be concluded that the
NDBI is increasing at a tremendous rate in KCC areas over the time.
The linear regression between LST and NDBI and the trend analysis in Figure 10 represents the rise of
LST with the increase of NDBI value over the time. The value of coefficient of determination, R2= 0.9867
in 2001(Figure 10 A) describes the strong responsive relationship between LST and NDBI. The
transformation of other land cover types in buildup areas has influenced to make the relationship stronger
in the year 2019. The value of R2= 0.9979 in the year 2019 (Figure 10 B) indicates the strongly significant
relationship between LST and NDBI. The coefficient of determination in the Figure 6, 8 and 10 suggests
that LST responded mostly with the change of NDBI value (as the relationship between NDBI and LST
found most significant) i.e. the increase of buildup area is mostly responsible for the increase of surface
temperature in the study area during the study time period.
3.5 LAND SURFACE TEMPERATURE(LST) CHANGE
The surface temperature change analysis shows that the average land surface temperature across the study
area has increased by 3.430C during the study period at the rate of 1.7140C/decade. Table 4 shows the
increase of the minimum, maximum and average temperature of different land cover during the study
period. The derived LST value range for each ward during the study period illustrated in Figure 11 shows
the clear increase of LST of each ward. The lowest LST increased by 1.710C, 1.850C and 0.740C at Ward
No 9, 14 and 16, respectively. The reason for this higher increase of LST value at ward 9 and 14 is that
the buildup area has increased mostly at these wards. LULC change direction analysis in Figure 4
demonstrated the declination of a huge percentage of ecofriendly land cover types such as agricultural
land, waterbodies, and vegetation land cover areas during the study period. This has adversely affected
the environment of the study area and influenced in the alternation of surface heat retention capacity. As
the buildup area was increased and the total amount of ecofriendly land cover types was decreased, the
percentage of area with higher LST value has increased. Such increase of surface temperature will reduce
the fertility of agricultural land and will increase soil salinity of the study area.
Table 4. Observed LST in different land cover in the year 2001 and 2019
Land use
LST in 2001 (OC) LST in 2019 (OC)
Minimum Maximum Average Minimum Maximum Average
Buildup Area 25.40 27.52 26.24 28.35 31.24 30.83
Vacant Land 24.55 26.67 25.80 26.19 30.42 29.95
Agricultural Land 23.68 27.10 24.85 25.45 28.18 27.39
Vegetation Land 23.68 26.67 24.83 25.83 30.01 28.23
Waterbody 23.68 27.93 26.16 25.40 30.42 28.88
The derived LST value range (higher value and lower value) in Figure 11A, 11C, 11D shows that the
range of observed LST in each ward was different in 2001 and the variation is also remained in the year
2019. As, the total amount of vegetation and waterbody areas has increased at ward 16 (Table 3), the
higher value of LST in this ward has decreased (28.760C to 28.700C) (Figure 11 E, F). The average LST
increased from 25.800C to 27.150C in Ward No 9, 26.840C to 27.230C in Ward No 14 and 26.870C to
27.120C in Ward No 16. The total amount of areas with higher LST value has increased mostly at ward
no 9 (Figure 11 C, D) and at ward 16 (Figure 11 E, F). Though the distance between these wards is very
less but the range of LST value varied in these three wards. Figure 3 and Figure 11 reveals that LST has
responded variedly in each ward as the LULC change also varied in each ward.
107
Figure 11. Land Surface Temperature map of KCC (A) Ward No 9 in 2001 (B) Ward No 9 in 2019 (C) Ward No 14 in 2001
(D) Ward No 14 in 2019 (E) Ward No 16 in 2001 and (F) Ward No 16 in 2019.
Table 5 shows that average LST of buildup areas increased mostly by 4.590C and the average LST of
vacant land, waterbody and agricultural land cover increased from 25.800C to 29.950C, 26.160C to
28.230C and 24.850C to 27.390C respectively (Table 4). Destruction of waterbodies and the rise of
atmospheric temperature has influenced in the increase of water surface temperature. The rise of
minimum, maximum and average LST in buildup areas land cover during the study time period and
regression analysis between NDVI and LST, NDMI and LST, NDBI and LST demonstrated that increase
of buildup area is mostly responsible for the increase of Land Surface Temperature in the study area.
Table 5. Nature of average surface temperature increase in different land cover during the study period
Land Cover Type Increased LST (0C) Avg. LST increasing rate (0C/year)
Buildup Area 4.59 0.229
Vacant Land 4.15 0.2075
Agricultural Land 2.54 0.127
Vegetation Land 3.4 0.17
Waterbody 2.72 0.136
3.6 LST DYNAMICS IN RESPONSE TO LULC TRANSFORMATION
Surface temperature changes during the study year in different land cover types was determined across
the study area. Increase of LST has been observed in all types of land cover transformation except
transformation of LC into waterbodies. Surface temperature has increased mostly due to the increase in
buildup areas (Table 6). Conversion of waterbody into build up area has increased surface temperature
mostly by 3.230C with a rate of 0.1610C/year (Table 6).
108
Table 6. Surface temperature change in response to land cover change in the study area.
Land Cover
Transformation
Avg. Temp in
2001(0C)
Avg. Temp in
2019(0C)
Avg. LST
Change (0C)
Increasing Rate
(0C/year)
Waterbody to Buildup 25.91 29.14 3.23 0.161
Waterbody to Vacant 24.83 27.88 3.05 0.153
Waterbody to Agricultural 25.19 27.01 1.82 0.091
Vegetation to Buildup 26.67 28.95 2.28 0.114
Vegetation to Agricultural 24.37 26.49 2.12 0.106
Vegetation to Vacant 25.57 27.18 1.61 0.081
Agricultural to Buildup 27.03 30.06 3.03 0.151
Agricultural to Vacant 24.54 27.40 2.84 0.142
Agricultural to Waterbody 26.01 25.55 -0.46 -0.023
Vacant to Buildup 26.89 29.93 3.04 0.152
Vacant to waterbody 26.23 25.84 -0.39 -0.020
Vacant to Agricultural 25.88 27.91 2.03 0.102
The analysis shows that the surface temperature has changed mostly due to the conversion of vegetation,
agricultural land and waterbody into buildup areas which is also found in the NDBI and NDMI value
change direction analysis and their trend analysis. Transformation of vacant land and agricultural land
into waterbodies has decreased the surface temperature by 0.390C and 0.460C, respectively. The analysis
shows that the surface temperature has risen in response to various changes in land use and land cover in
different areas of the study area. The overall land cover transformation rate across the study area is
indicating a huge increase of surface temperature in the future.
4.0 CONCLUSION
The study reveals a drastic land use and land cover transformation of agricultural land and waterbodies
in the study area. Unplanned urbanization has resulted in the increase of 376.59acres (17.82%) of buildup
area and destruction of 227.27 acres (10.75%) of agricultural land and 110.73 acres (5.24%) of
waterbodies in the study area in just 18 years. This high increasing rate of buildup area has influenced in
the unexpected change of NDMI, NDVI and NDBI indices values. Though the three wards are located
side by side, but the LST and NDMI, NDVI, NDBI indices values have changed differently as the land
cover transformation type varied in each ward. The Most distinct observation was that the surface
temperature has increased mostly in those areas where waterbodies and vacant land has transformed into
buildup areas. Overall study demonstrates that the land covers are not changed in a planned way in
everywhere in Khulna city. As a result, environmental changes in microlevel are taking place very subtly
such as the rapid increase of surface temperature, atmospheric temperature, and changes in landscape.
This unplanned land cover type conversion into buildup areas has increased the average LST by 3.430C
with the rate of 1.7140C/decade across the study area. Such land cover change at micro level are causing
many other adverse reactions to its environment, which in the macro level such as city level is often not
well realized. The valuation of environmental change at the micro level through this study will help to
aware people about the significance of planned land use change and researcher to focus on the root level
land use dynamics to mitigate the adverse effects of it on the environment.
5.0 ACKNOWLEDGEMENT
The authors gratefully acknowledged the EarthExplorer -USGS as the source of satellite images and also
thankful to Professor Mustafa Saroar, KUET for giving the theoretical knowledge. Mr. Saif Siam, Mr.
Fazle Rabbi and Fahmida Yeasmin Sami helped in data collection. Anonymous reviewers are also
109
acknowledged for their crucial comments that have helped the authors to improve the manuscript
substantially. The authors are also grateful to the editorial team.
REFERENCES
[1] U. Shahzad, 2015. Global Warming: Causes, Effects and Solutions. Durreesamin Journal, Vol. 1, No. 4.
[2] M. N. Haque, M. A. Mamun, M. M. Saroar and T. K. Roy, 2019. Applica1`tion of “DPSIR” Framework to Assess
the Status and Role of Blue Ecosystem Services (BES) in Khulna City. Journal of Engineering Science (JES), Vol.
10, No. 2, pp. 49-60.
[3] P. K. Srivastava, T. J. Majumdar and A. K. Bhattacharya, 2010. Study of land surface temperature and spectral
emissivity using multi-sensor satellite data. Journal of Earth System Science, Vol. 119, pp. 67-74, 9.
[4] P. Dash, F.-M. Göttsche, F.-S. Olesen and H. Olesen, 2002. Land surface temperature and emissivity estimation from
passive sensor data: Theory and practice-current trends. International Journal of Remote Sensing, Vol. 23, No. 13,
pp. 2563-2594.
[5] J. Cristóbal, J. C. Jiménez‐Muñoz, J. A. Sobrino, M. Ninyerola and X. Pons, 2009. Improvements in land surface
temperature retrieval from the Landsat series thermal band using water vapor and air temperature. Journal of
Geophysical Research, Vol. 114, pp. 103-129.
[6] A. Gillespie, S. Rokugawa, G. T. Matsuna, J. S. Cothern, S. Hook and A. B. Kahle, 1998. A temperature and
emissivity separation algorithm for Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER)
images. IEEE Transactions on Geoscience and Remote Sensing, Vol. 36, No. 4, pp. 1113-1126.
[7] Z. Qin, A. Karnieli and P. Berliner, 2001. A mono-window algorithm for retrieving land surface temperature from
Landsat TM data and its application to the Israel-Egypt border region. International Journal of Remote Sensing, Vol.
22, No. 18, pp. 3719-3746.
[8] F. S. Kiran and B. U. Joshi, 2019. Estimation of variables explaining urbanization concomitant with land-use change:
a spatial approach. International Journal of Remote Sensing, Vol. 34, No. 3, pp. 824-847.
[9] A. Sánchez-Lugo, P. Berrisford, C. Morice and A. Argüez, 2018. Temperature [in "State of the Climate in 2017"].
Bulletin of the American Meteorological Society, Vol. 99, No. 8.
[10] S. Shahid, 2010. Recent trends in the climate of Bangladesh. Climate Research, Vol. 42, No. 3, pp. 185-193.
[11] M. A. Fattah, S. R. Morshed, M. N. Haque and S. Rahman, 2020. Land Use Changes and Its Impact on Environment:
A Case Study on Urban Area in Khulna, Bangladesh. International Conference on Civil Engineering for Sustainable
Development, Khulna, 2020.
[12] K. V. Singh, R. Setia, S. Sahoo, A. Prasad and B. Pateriya, 2015. Evaluation of NDWI and MNDWI for assessment
of waterlogging by integrating digital elevation model and groundwater level. Geocarto International, Vol. 30, No. 6,
pp. 650-661.
[13] S. Guha, H. Govil, A. Dey and N. Gill, 2018. Analytical study of land surface temperature with NDVI and NDBI
using Landsat 8 OLI and TIRS data in Florence and Naples city, Italy. European Journal of Remote Sensing, Vol. 51,
No. 1, pp. 667-678.
[14] A. Varshney, 2019. Improved NDBI differencing algorithm for built-up regions change detection from remote-
sensing data: an automated approach. Remote Sensing Letters, Vol. 4, No. 5, pp. 504-512.
[15] NASA, "Global Warming," 03 June 2010. https://earthobservatory.nasa.gov/features/GlobalWarming/page2.php
[16] B. Ahmed, M. Kamruzzaman, X. Zhu, M. S. Rahman, and K. Choi, 2013. Simulating Land Cover Changes and Their
Impacts on Land Surface Temperature in Dhaka, Bangladesh. Remote Sensing, Vol. 5, No. 11, pp. 5969-5998.
[17] D. R. Raja, 2012. Spatial Analysis of Land Surface Temperature in Dhaka Metropolitan Area. Journal of Bangladesh
Institute of Planners, Vol. 5, pp. 151-167.
[18] N. S. Parvin and D. Abudu, 2017. Estimating Urban Heat Island Intensity using Remote Sensing Techniques in Dhaka
City. International Journal of Scientific & Engineering Research, Vol. 8, No. 4, pp. 289- 298.
[19] M. B. Mia, R. Bhattacharya, and A. Woobaidullah, 2017. Correlation and Monitoring of Land Surface Temperature,
Urban Heat Island with Land Use-Land Cover of Dhaka City Using Satellite Imageries. International Journal of
Research in Geography (IJRG), Vol. 3, No. 4, pp. 10-20.
[20] S. Ara, M. A. Islam, and S. Showkat, 2016. Effect of land-use intensity on surface temperature: A study on Chittagong
city corporation area," in IEEE Xplore Library.
[21] A.-A. Kafy, A.-A. Faisal, M. M. Hasan, M. S. Sikdar, M. H. H. Khan, N. Rahman, and R. Islam, 2019. Impact of
LULC Changes on LST in Rajshahi District of Bangladesh: A Remote Sensing Approach. Journal of Geographical
Studies, Vol. 3, No. 1, pp. 11-23.
[22] G. Grigoraș and B. Urițescu, 2019. Land Use/Land Cover changes dynamics and their effects on Surface Heat Island
in Bucharest, Romania. International Journal of Applied Earth Observation and Geoinformation, Vol. 80, pp. 115-
126.
[23] H. Xu, 2006. Modification of normalized difference water index (NDWI) to enhance open water features in remotely
sensed imagery. International Journal of Remote Sensing, Vol. 27, No. 14, pp. 3025-3033.
[24] W. C. Snyder, Z. Wan, Y. Zhang and Y.-Z. Fenz, 1998. Classification-based emissivity for land surface temperature
measurement from space. International Journal of Remote Sensing, Vol. 19, No. 14, pp. 2753-2774.
110
[25] S. Guha, H. Govil, A. Dey and N. Gill, 2018. Analytical study of land surface temperature with NDVI and NDBI
using Landsat 8 OLI and TIRS data in Florence and Naples city, Italy. European Journal of Remote Sensing, Vol. 51,
No. 1, pp. 667-678.
[26] H. West, N. Quinn, and M. Horswell, 2019. Remote sensing for drought monitoring & impact assessment: Progress,
past challenges and future opportunities, Remote Sensing of Environment. Vol. 323, p. 111291.
[27] M. Sahana, R. Ahmed and H. Sajjad, 2016. Analyzing land surface temperature distribution in response to land
use/land cover change using split window algorithm to land use/land cover change using split window algorithm and
spectral radiance model in Sundarban Biosphere Reserve, India. Modelling Earth Systems and Environment, Vol. 81,
No. 2.
[28] A. -. A. Kafy, A.-A. Faisal, M. M. Hasan, Abdullah-Al- Faisal, M. Islam, and M. S. Rahman, 2020. Modelling future
land use land cover changes and their impacts on land surface temperatures in Rajshahi, Bangladesh. Remote Sensing
Applications: Society and Environment, Vol. 18, No. 2.
[29] S. M. T. Rahman and A. Kabir, 2019. Factors influencing location choice and cluster pattern of manufacturing small
and medium enterprises in cities: evidence from Khulna City of Bangladesh. Journal of Global Entrepreneurship
Research, Vol. 9, pp. 61-87.